Very slow deployment with LabVIEW 2011 vs. 8.6 on cFP 2120

[MarkCG]

Hi all, 

I have been developing with LV for about 4 years now, and I have been working with Fieldpoint controllers for a couple of months.

I took over a project written in LV 8.6. Deploying to a cFP was a fast process, about 2-3 minutes at most, after rebooting the cFP. I then decided that I was tired of working in the 8.6 dev environment and upgraded to LV2011, and updated the real-time engine on a cFP to see how things worked. Unfortunately, it now seems that it takes more like 20 minutes to deploy the application I'm working on. Very irritating! I didn't do anything like make the application bigger, I actually made it smaller. No property nodes anywhere either, I saw a that reported those caused slowdowns. I did do a full install of LV realtime on the cFP, maybe it's one the many unused services (Epics server? Any of the myriad web services?) installed?

Any ideas why this might be? 

Regards,

Mark G

[Andrew_B]

Hello Mark,

This is Andrew Brown, an Applications Engineer. It is certainly not expected behavior for your application to take longer to deploy using LabVIEW Real Time 2011. Rather, the deployment process should be faster according to this article. To gather more information about the issue, I would appreciate if you could clarify your system:

1. I assume you are working with FieldPoint 6.0.9 now. Did you upgrade this driver's version as part of upgrading LabVIEW? If so, what version were you working with before?

2. Are you using any additional toolkits with your LabVIEW program? (Statechart, Mathscript RT, etc.)

3. Does this deployment slowdown happen only with this program, or does it also happen with examples from the NI Example Finder?

Regards,

Andrew Brown

[MarkCG]

Hi Andrew, 

Thank you for your help. I did have LVRT 8.6 v6.0.4 on that cFP before and installed LVRT 2011 v6.0.9 on it. It is very difficult to use, with constant "waiting for target to respond" messages. I will likely try to reformat the drive. The deployment appears fast at first, but it soon slows down to deploying a VI every 5 to 30 seconds.

I also have another cFP, exact same RT controller (cFP 2120) and modules,  that has LVRT 8.6 v6.0.4 and deployment to that is very fast. I am afraid of installing LVRT 2011 until I can assure myself I can un-install it. 

I am not using any additional toolkits.

 I will perform the test using example VI as soon as I have time. 

Regards, 

Mark G

[MarkCG]

I am starting to think that there may be something wrong with the controller. I installed both LVRT 8.6 with 6.0.4 driver, and LVRT2011 with 6.0.9 drivers. Now it seems the only way the unit responds when pinged is when it's in safe mode. Will try another controller.

[MarkCG]

Andrew, 

I tried the experiment on a completely different cFP unit now, another cFP 2120. Completely different router. With LVRT 2011 6.0.9 (minimal) installed. I first deployed my project. This took about 20 minutes (!). I have the deploy log saved so I can send it to you. Did not run the main VI. 

I then deployed an example VI "RT Global Communications". VI deployed successfully, but after deploy complete received the "waiting for target to respond message". Restarted the cFP. Status light is OFF, no blinking, nothing. No response to ping. 

Rebooted cFP in safe mode with dip switch, status light blinks three reds. cFP responds to ping.  

Use find remote systems in MAX. Does NOT come up in MAX at all. MAX hangs, closed via task manager. 

Ping cFP, "reply from 192.168.1.101: Destination host unreachable" or "request timed out"

Reboot again with safe mode switch on. Status light blinks three reds. Ping cFP: "PING: Transmit Fail. General Failure".

Ping again "reply from 192.168.1.101: Destination host unreachable" or "request timed out".

Reboot with safe mode and with "reset IP" switch on. 

MAX finds cFP. Select "DHCP or link local" on network settings, save, unit restarts. Left IP reset switch on. Mistake. MAX finds unit, IP still set to 0.0.0.0 as expected. Configure static IP:

IP: 192.168.1.101

Gateway: 255.255.255.0

Gateway: 192.168.1.1

DNS: 0.0.0.0

Save in max, cFP reboots. MAX error: "there was a problem retrieving setting from target.... enter info code MaxServerUnknownError".

Try reset IP one more time. Safe mode still on. this time select "DHCP or local". Flip ip reset switch back. Reset. MAX cannot find.

I'm pretty much stuck. My project deploys just fine and fast when running on targets running LVRT 8.6. Doing so on LVRT 2011 seems to brick the unit. Please help....

Regards,

Mark G

[Andrew_B]

Hello Mark,

I understand that you have been working with another Applications Engineer on the issue of detecting your cFP 2120. At this point, I understand that you can detect and work with this compact FieldPoint device using MAX and LabVIEW. Is this your current status?

If so, I would still be very interested in the deployment time of examples from the NI Example Finder. Could you please try one of these examples and let me know the resulting deployment time?

Regards,

Andrew Brown

[MarkCG]

Just wanted to update the thread. Andrew took my code, set up his own cFP 2120 system and confirmed that deployment in interactive took about 12 minutes with LV2011.Take about 2 minutes in LV 8.6.1 .

I did resolve communication problems, which are unrelated. This was indeed strange. The ethernet hub I was using was plugged into a number of other computers and who knows what else. It changed its gateway IP  from 192.168.2.1 to 192.168.168.1.1 and back to 192.168.2.1 entirely on its own. In the meanwhile, it acquired / lost / reacquired a connection to the intenet. I know this from looking at my project file and the series of events. 

Thank you Andrew for helping me out with this issue. I do appreciate it.

[Andrew_B]

As a further update, I tested deployment time to the cFP-2120 with different versions of LabVIEW using Mark's program. The results are as follows:

LabVIEW 8.6: 1 - 3 minutes 
LabVIEW 2010: 1 - 3 minutes 
LabVIEW 2011: 10 - 20 minutes 

This points to using LabVIEW 2010 as an viable option to reduce deployment for this specific situation. Deployment time with LabVIEW 2011 is faster in the general case, but in this edge case ( cFP-2120 with a large program) a previous version of LabVIEW appears to improve deployment time.

[Timmar]

I have had similar slow download behavior,

I am modified my code to use Labview Classes/Objects with override methods and I suspect that this has greatly increased the download time.

I am using LV 10.0.1 on a CFP 2220 with cFP ver 6.0.8

[MarkCG]

That's interesting. I'm going to try using LV2010 as Andrew suggested. I'm not using any LVOOP at all in my project, so hopefully I won't have the same problem.  

I'm going to gripe here for a minute, so please excuse me....

I would like to state that although this particular issue is in a way, a "corner case" it doesn't look good wrt respect to LabVIEW being used for "real" industrial control systems. When a company spends the money to buy 20+ cFPs and all the 6-7 modules per chassis, they see this as a capital investment. They expect it to be supported for 15+ years.  They see it as a PLC, essentially. 

You could argue that one should keep using LV 8.6, but really, is that realistic? What about 4-5 years from now? The cFPs will still be in working condition. Technically, LV 8.6 isn't even supported on win 7.. The cFP series is being sold on the NI website now, it should be expected that the latest version of LV should work with it.